File: MOUSE_RG.TXT        Guy Dunphy                3/7/1998



                          MOUSE RAGE!



Road rage. We've all heard of that, and maybe felt a bit of it ourselves,

simmering in some crawling, exhaust-fogged example of commuting madness.



We take this business of moving ourselves from one place to another pretty

seriously, don't we? So you'd think we'd put in the effort to get it right.

Not likely! The forces of commerce, fashion and conformity beat common sense

any day, and keep making a mess of our hopes and lives. To the point where

I can sympathise with the desire to put a bullet through the nearest car.



But the roads are not the only place where modern life can fray tempers to

the point of murder. Rage has come to the desktop: Mouse Rage!



Today I put a bullet through my mouse. Shot it dead - metaphoricly speaking,

since its owned by the company I work for. Close enough though. Its sitting

now on the far corner of the desk, cord coiled tight round its throat, with

a yellow post-it note attached, saying 'This mouse is shot!' Hopefully by

next Monday, someone will have removed the corpse.



With the roads, we can blame any number of causes for our frustrations:

the government, road planners, car makers, the frantic drives of society,

or the driving habits of the recently arrived. Even, if we are honest,

ourselves. Its hard to say where the true blame lies, other than that

human folly will be in there somewhere.



But with mice, the blame is clear. Its _all_Microsoft's_fault_! Not only

are Microsoft collectively the biggest pack of bastards, liars and con

artists on the planet, to judge by the crap software they flog off for

exhorbitant prices, but their dammed mice are crap too. Don't work properly,

never did, and never will.



For moving a cursor about a screen with any delicacy of control, MS mice,

such as the 'Serial Mouse 2.0' I shot today, are pathetic failures. Even

new, these mice don't work well. Without a speck of dust inside them, and

perfectly clean rollers, for some reason the ball seems to stick as you

make small, slow movements of the mouse. This makes the mouse useless for

any fine detail work like digital artwork, and extremely frustrating even

when simply trying to get an editor cursor into approximately the right

position in a text file. Mouse mats are to some extent an attempt to fix

this problem, by increasing the friction with the static surface. This

does help get the ball rolling, but since the mat tends to creep, and

its surface deflects a little, very fine control is still difficult.



Then it gets worse. Due to the materials MS chose for the ball and rollers,

MS mice seem to act like vacuum cleaners for dust and grime, scooping it

up from the desk surface as it adheres to the ball, then packing it solidly

onto the rollers in a lumpy ring of congealed, impacted mouse-shit. Hence

the regular filth-picking ritual known to most MS mice users, where one

struggles to dislodge the ring-of-hesitancy from the mouse's inner privates

with a probing fingernail.



This 'stickiness' was the reason I shot my mouse today. I finally got sick

to death of the annoyingly jerky, approximate control it gave of the cursor.

I went and bought myself a _real_ mouse. Or tried to. More on this later.



So mechanically, MS mice are cripples. They are also severely deficient in

the buttons department - MS mice have exactly two thirds as many buttons

as are optimal for a sensible User Interface pointing device.

Consider for a moment how one holds a mouse. Thumb and little finger are

used to hold the mouse body on its sides. That leaves the three 'middle'

fingers sitting on the top surface of the mouse, uninvolved in the business

of dragging the mouse around. This alone should be fairly suggestive.

Then consider what actions one might want to trigger using single button

clicks, without moving the mouse. I won't bore you with an exhaustive

analysis of many different situations, but lets look at just one: moving

around in a text file. At the most basic level, there are really exactly

three possibilities: 'move up relative', 'move down relative', and 'go to'.

So duh, how many buttons should be available for this?



This same sort of thing is repeated over and over in different cases: one

button is seriously limiting, two is still not enough, three is a good fit,

and more than three don't give significant gains in functionality.

Considering both the ergonomics of how we hold a mouse (three fingers

available), and the control needs of most sorts of WIMP UI or application

(most need at least three buttons, but usually not more), the obvious

conclusion is that three button mice are best.



Historically, what happened was this:-

 - Logitech and a few other mouse makers got it right, with three buttons.

 - Apple, being the idealogues they were, and obsessed with 'simplicity',

   especially the superficial appearance of it, chose to use just one button. 

 - Microsoft, of course, had to be different. Market differentiation, etc.

   So they thought hard for several microseconds, then picked a number at

   random between three and one.

   Thus demonstrating once again their fundamental stupidity, and damming

   millions of computer users worldwide to an inferior pointing interface.



Those early Logitech mice were beautiful. Their nylon rollers never seemed

to pick up dirt from the ball, so never needed cleaning. The ball never

shows the slightest sign of sticking, with even the slowest, tiny movement

being translated perfectly to the cursor. The rectangular body (ie the old

Logitech C7 mice) was a good weight, and the rim along the top of the sides

meant it was very easily held positively by a loose thumb and little finger.

Also, it had something _none_ of the so called 'ergonomic' modern mice have-

a finger rest area to the fore of the buttons. So it wasn't necessary to

keep the extensor tendons of the hand tense all the time, just to avoid 

accidentally clicking buttons. This and the easy side grip meant that the

thing was perfectly comfortable to use, for any length of time. Also, when

you did rest your fingers on the buttons, your fingers were in the most

natural, 'half curled' position, not nearly fully extended as with most

newer mice.



It may have looked a bit angular (actually I prefer that), but these three

button mice were (and are) the prefered choice for most professional CAD

software. Where practical usability (and three buttons) mattered more than

ideology, early Logitech mice were supreme. I still have two at home, and

prefer them to any other mouse I've ever used. About the only way in which 

these mice could be improved, would be to replace the ball-and-rollers

system (which does tend to stick if there's grit on the table), with the

alternative of two axially tilted wheels, that contact the desk surface

directly. Mice like this (I have one, by Keytronic) are totally immune to

dust, hairs and grit. Unfortunately, I have yet to find one such that comes

with workable driver software, or which emulates an adequate range of mouse

serial protocols - particularly Logitech's. But perhaps that is a software

patent related problem, rather than mere incompetance on the designer's part.



Lets come back to MS Windows for a moment, and consider the ergonomics of

its User Interface. Hands up those who think it sucks.  Hmm... I thought

you'd think so. Has anyone here considered _why_ it sucks? Or how it might

be improved? Ummm OK, silly question, of course you all have many, many

suggestions. But may I point out that there is one, fundamental cause of

much of Window's klunkiness - Microsoft's dependence on a two button mouse.



Take the basic actions one might want to do with an object. These are:

a) Select it. b) Query it. c) Act on it (includes 'cause it to act'.)

With three buttons, there would be no need to use double clicking, or at

least, not as a primary command.

Then take the Windows scroll bars (sorry things that they are.) With three

buttons, a _much_ more efficient scroll bar functionality is possible:

 - Left:    Move up in view by amount proportional to position of click

            on the bar.

 - Middle:  Move to absolute position in view, based on the bar representing 

            the complete available range, and the click position the desired

            location in that range.

 - Right:   Move down, scaled same as for Left.

This is _so_ much simpler and more convenient in practice, it has to be tried

to be believed. I have a text editor written in the mid eighties with scroll

bars that work this way. The best thing is that you can move forward and back

in the file, in controlable but equal steps, without moving the mouse.

And go to any exact location in the file with a single click. It makes me

weep to go back to Windows editors and bars, which are pathetic by comparison.

I won't detail the many things that are wrong with the design of MS Windows

scroll bars, except to say that they are overcomplicated and tedious to use.



Naturally, there's more than mere bad design to the Windows scroll bars.

With Microsoft, its not surprising to smell the taint of deliberate crippling

in the air, and this is no exception. In this case its a 'bug' thats really

annoying, has existed for years, and has been seen by almost everyone at one

time or another. Its the 'scrolls out of control' bug, where you click on

the bottom arrow, the screen starts scrolling up, you see what you want,

you release the mouse button.... and the window _keeps_scrolling_. Somehow

the 'button released' event got lost. You have to reclick to stop it.



I've wondered for at least two years why this 'bug' was not being fixed.

Then I found the answer - the new, Microsoft 'scroll mouse' - the one with

the little roller where a middle button should be. Ah, I see. A marketting

ploy. Get people so fed up with an existing interface, that they'll buy

anything that offers a solution - even another crap Microsoft product.

A stupid hack, to fix a stupid problem. With three buttons, easy, rate

controlled auto-scrolling would be simple to arrange.





Anyway, today I went down to the local computer superstore, to buy me a

new mouse. A good one, and therefore, one not by Microsoft.

It was a truly enraging experience. All I wanted was a plain, three button,

plug into a serial port, mouse. And hopefully software compatible with the

logitech mouse interface, since thats what my favourite old editor likes

to see. Oh, and it must work OK with Quake.



And what do I find? A wall full of mice, but all either deformed into weird

and fantastic shapes, or having only two buttons. Whats with all these 

crazy shapes? This is not a science fiction movie set, this is my desk.

At the very least, I want the top of my mouse to be reasonbly close to

horizontal, but most of these nightmares are sloped like they want to shed

snow. Seems like the marketdroids and styloonies have completely taken over.

Even Logitech has backslid, and now sells a two button mouse! Horrors.



There was one almost normal mouse there, in a slightly battered looking box.

A three button Logitech, "for the professional on the go". Which meant it

was for use with a laptop - and hence had a whole two feet of cable, and

was priced for the laptop market, ie exhorbitantly. Pay more for less, what

a deal. I took it.



At least while there I had the satisfaction of overhearing a suit discussing

the purchase of Windows 98 with a salesman, since he thought he 'needed to

upgrade to Internet Explorer'. So I volunteered the information that the

Netscape browsers are free for the download (and better), how MS IE is buggy,

insecure and non-Java compliant, and a quick review of the many sins of

Microsoft, including that their web servers deliberately respond slower to

requests from Navigator, than from IE. He went off, one less sucker to

Microsoft's browser scam.



Back at work, after digging up a serial extension cable to reach the back of

the tower case, happily got back to hacking the current pile of code. And at

_last_, I can put the cursor right.... _there_, first time.



For the future though, I can see that mice will be a problem. For one thing,

I'll soon be adding a few more machines to my Quake local LAN, and each one

will need a mouse. And what if my trusty, old Logitech 'boxy' mice break?

I don't want to be forced to use the fancy, overstyled excuses for mice,

really I don't.



So can anyone help me? What I want, is several working, old Logitech mice,

especially the 'C7' model. They look like a flat rectangular box, with part

of the top surface sloping forward at a shallow angle. And 3 buttons, natch.

I'd like to lay in a stock, to last the rest of my life - or till everything

breaks on Jan 1st, 2000, whichever is sooner. :-) Maybe about ten of them.

Even broken ones will be gladly accepted, since I don't mind swapping bits

to get one or two working out of several.



If you feel charitable to send one or a few free, then thanks _very_ much

for your generosity, and I'll reply post with repayment of your mailing cost.

Alternatively, if you've got a drawer-full of them, reply email me and we'll

talk about price.

Please address post to

    Guy Dunphy

    PO Box 262

    Panania, 2213

    Sydney,  NSW 

    Australia.



----------------------------------------------------------------------

3/11/1998 More Mouse Misery



Another soul fed up with a Microsoft 'Serial Mouse 2.1A' gave it to me in

disgust - which presented an opportunity to do an investigative rodent

dissection. The question is: how exactly does MS manage to make these mice

so very *bad*? And coincidentally, how do they get to revision 2.1A and

still not have it anywhere near right?



Well, the answer is that Microsoft actually manage to get *everything* wrong,

and here is the medical record to prove it.



Long term prior illness: This mouse reported to have had 'skipping disease',

where while sliding it along, for no reason the cursor just freezes. Then

recovers when the mouse is jiggled a bit in any new direction. This naturally

made the mouse very frustrating to use.



Ultimate symptom of death: It suddenly just stopped tracking completely, and

no amount of percussive desk therapy could restore it.



Initial examination of corpse: The mouse appeared intact, virtually brand

new. On testing it was, indeed, dead in one axis, but working in the other.

Bottom cover and ball removed, rollers visually inspected and found to be

lint & scum free. Closer inspection revealed one roller to be slightly

'lifted' from its bearings, and a gentle poke in the right direction

settled it back in with a small 'click'. Now the mouse works, but a brief

trial reveals it to be still infected with skipping disease.



After removing the top cover (the one screw is hidden under a stick-on

slip foot) I could actually watch the ball rotate as the mouse is slid

across the, er... operating surface. After a little trial and error, I

found a particular sweeping, curved drag path that would fairly reliably

result in loss of cursor tracking. Looking closely, it was clear that

the ball was actually lifting up and to one side at these moments, thus

losing contact with one or both of the X/Y sense rollers.

So why does it do that? Other mouse manufacturers don't seem to have any 

problem keeping the ball in the right place. Thats what the spring-loaded

45 degree roller is for - to keep the ball pressed against the other two 

rollers.

But in this MS mouse, the 'spring' is a bent bit of wire, one end carrying

the roller, the other end (3cm away) being flimsily click-mounted in some

retaining lugs on the plastic base. One factor at work here is that the

roller end of the spring wire 'drifts around' (up and down mostly) as the

ball rotates in various directions. Also the roller drifts along the wire

between two fixed stops. With some combinations, the roller wedges into

an end-stop at a bit of an angle, and seems to get more frictional drag

than is healthy in mouse guts.

But watching this happen over and over, it didn't seem like quite the full

explanation. Sometimes with the roller rubbing the ball would shift 'off

roller', and sometimes not. A bit less repeatable than the vagaries of 

friction could explain.



While fiddling with this, I happened to take the ball out again and put

it on the desk. It rolled... back and forth. Huh! This thing has an off-

center 'center of gravity'.  Now, you may be aware that these balls are not

plain rubber, but a steel ball embedded in a rubber skin. Obviously, you'd

want the steel ball to be in the exact middle of the rubber, if you wanted

to avoid intermittent directional tracking errors.



Cutting the MS ball open, I found a badly off center steel ball. The

thickest point of the rubber skin was 2.1 mm, and the thinnest 1.3 mm.

So the mouse was skipping when a conjunction of badly mounted tension

roller, and off-center ball, acted to allow the ball to drift out of

position, and lose contact with the sense roller(s).



Conclusion: This is a gross and intrinsic design flaw. It is not repairable,

and is likely to affect all units of the same model number to a greater

or lesser extent (depending on the ball mass eccentricity.)

MS should be forced to recall the entire population of these mice, and

issue replacements (with a better engineered model.)